JPS6034318B2 - image spatial frequency analyzer - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an image spatial frequency analyzer that configures spatial frequency components in the horizontal scanning direction of a television image into an image in a form corresponding to an input image and displays the image.

Conventionally, an image is input by a television camera, and a spatial real-time correlation calculation process is performed using an electro-optical method to obtain a horizontal scanning direction correlation image of the image, which is then displayed on the screen of a receiver. There is a correlation device.

In such a device, the two signals used in the correlation calculation are both processed signals such as television image video signals, so even if the correlation output image is detected, the spatial frequency components contained in the television image cannot be detected. Have difficulty. That is, in the autocorrelation output image, only the periodicity of the input image can be detected, and in the cross-correlation output image, only the degree of similarity (degree of overlap) with respect to the specific target signal can be measured. Here, the authors first discuss the 197 single king 3 of the journal of the Institute of Electronics and Communication Engineers.
Month Vol. 621A, No. 3 “Ultrasonic light modulator and TV
We will describe the construction method and operating principle of the conventional image correlation device, which is an example of the device presented in "Construction Method of Image Correlation Device Using a Camera").

FIG. 4 shows a schematic diagram of the configuration of a conventional device. This device focuses on the fact that the video signal of a commonly used television camera has a repetition period and frequency band suitable as an input signal to an ultrasonic optical modulator used in an electro-optic correlator. performing a one-dimensional autocorrelation calculation on the video output signal for each horizontal scan of the television image to be correlated by the correlator;
The results of this correlation calculation are constructed into an image on the screen of the receiver. Therefore, a horizontal scanning direction autocorrelation image of the input image can be obtained as an output image. In the apparatus shown in FIG. 4, an input image 13 to be autocorrelated is captured by a television camera 14 that performs bidirectional horizontal scanning using a triangular wave with a frequency of 7.87 pitches 2 generated by a bidirectional oscillator 15. ,
A synchronization pulse separation circuit 101 is added to separate the obtained composite video signal into a video signal and horizontal and vertical synchronization pulse signals.

The video signal separated here is sent to the video signal pulsing circuit 9 for the purpose of satisfying the conditions for optical correlation calculation described later, and is converted into a binary pulse video signal representing black and white binary values by comparison with a set threshold value. converted. This binary pulse video signal is applied as an external modulation signal to a pair of sine wave oscillators la and lb which are capable of pulse amplitude modulation. The oscillator oscillates at the resonance frequency of the ultrasonic transducers placed in each of the ultrasonic light modulators 5a and 5b, and the number of oscillations is amplitude modulated by the binary pulse video signal within the oscillator and output. be done. This oscillator output signal is applied to the ultrasound transducer and radiated into the ultrasound light modulators 5a, 5b as spatial ultrasound signals. Next, the optical correlator will be explained.

A laser beam 4 is used as the light source of the optical system, and after the laser beam is expanded by a lens, it is made incident on the light incidence windows of the ultrasonic light modulators 5a and 5b. A pair of the ultrasonic light modulators 5a
, 5b have opposite directions of ultrasound propagation, and
The ultrasonic wave propagation direction and the optical axis are arranged so as to intersect at right angles. The light beams that have passed through the ultrasonic light modulators 5a and 5b are partially phase modulated by the spatial ultrasonic signals in the modulators, and are converged by a lens to form a ridge on the focal plane. A photodetection filter 6 is disposed on the focal plane, and detects the first-order diffracted light generated by the resonance frequency wave of the ultrasonic transducer. The light that has passed through the photodetection filter 6 enters a photoelectric converter 7 and is converted into an electrical signal. Generally, when the amplitude of the ultrasonic signal is small, the amplitude value and the first-order diffracted light amplitude value are in a proportional relationship, and the photoelectric converter output current value, which is proportional to the light amount value which is the square of the light amplitude value, is the ultrasonic signal amplitude It is known that it is proportional to the square of the value. The principle of optical correlation calculation will be explained in detail later. Now, the synchronous pulse separation circuit 10
The synchronization pulse separated by the horizontal synchronization time delay circuit 11
and is added to the mixer 12 after being subjected to a time delay. This mixer 12 adds a delayed synchronization pulse to the correlation output signal obtained by the photoelectric converter 7 to create a composite video signal, and the mixer output signal is configured into an image in the image receiver 16. By setting the delay time of the horizontal synchronization time delay circuit 11 to approximately 1'2,32 ys of the horizontal synchronization time, it is possible to display the autocorrelation image of the input image 13 at the center of the screen of the image receptor 16. It is. Next, the principle of optical correlation calculation will be described.

Basically, a video signal obtained by bidirectional horizontal scanning, the scanning direction of which is reversed for each scan, is spatially inverted, delayed, and shifted (moved) by a pair of ultrasonic optical modulators 5a and 5b. To obtain a cross-correlation function between temporally adjacent video signals, that is, a vertical relationship on a TV screen, by approximating the cross-correlation function to an autocorrelation function by converting it into a signal and performing an integral calculation using a lens. It is. As shown in Fig. 5, the input image is a binary image P (Q, 8), and the video signal obtained in the nth horizontal scan after being taught from the vertical synchronization pulse is f
Let it be (Q, n). As shown in FIG. 6, the ultrasonic signal that has been shaped into a binary pulse video signal and amplitude-modulated by f(Q, n) is applied to an ultrasonic transducer in an ultrasonic light modulator. Since the ultrasonic light modulators 5a and 5b are arranged so that the directions of ultrasonic propagation are opposite to each other, the ultrasonic signals in the ultrasonic light modulators are U, (x, n), U2 (1× ,n)
It can be written as The ultrasonic light modulator 5a is made longer in the ultrasonic propagation direction than the ultrasonic light modulator 5b, and the ultrasonic light modulator 5a has a longer length in the ultrasonic propagation direction than the ultrasonic light modulator 5b.
The time it takes for x, n) to reach the laser beam east is U2(1×
, n) reach the same time, one horizontal scanning time is 63.5
There will be a delay of one second. Therefore, the number of ultrasonic signals that overlap within the laser beam is U, (x, n) and (1 x, n + 1).
The superposition of two signals within the optical system is expressed by a product operation, and the convergence effect by a lens is expressed by an integral operation. Also, U
, (x, n), U2 (1 x, n+1) are moving inside the ultrasonic optical modulator at the ultrasonic velocity v, so only the diffracted light generated by the ultrasonic signal is detected by the photodetection filter. Evening 6
By detecting it and photoelectrically converting it, the following correlation output C(t,i) can be obtained as a current value. '1} In the formula, d represents the width of the Caesar luminous flux, and g(x, i) represents U(x,
n), that is, a function obtained by linearly converting Q of the video signal f(Q, n) to x.

If the number of bidirectional scans of the TV camera is sufficiently large compared to the size of the image, the following relationship holds true: Assign to and get.

The glue equation represents the autocorrelation function of g(x,i), ie, f(Q,i). Since both C(t, n) and f(Q, n) are functions of the number of horizontal scans n, C(t, n) is
If an image is constructed using a synchronization signal of (t, n), a horizontal scanning autocorrelation image of a binary image can be obtained. However, in order to perform the glue method and obtain independent correlation outputs for each horizontal scanning video signal, a no-signal portion of equal length is required before and after each video signal, so it is necessary to The length of the signal is limited to 1/sa of the horizontal synchronization time. Further, in the optical correlator, the following square value correlation calculation is performed due to the square characteristic of the photoelectric converter 7. [4
} expression is equivalent to the correlation operation of expression '3}, g(x,
It is necessary to limit n) to a binary value of [1, 0], and specifically, the input image is limited to a black and white binary image, or the video signal is converted into a binary pulse using a threshold circuit. The above is an overview of the configuration and operating principle of the conventional device. An object of the present invention is to improve the above-described image correlation device, apply correlation calculations to detect spatial frequency components contained in television images, and display the detected spatial frequency components in the form of images.

In other words, the video signal of the input image obtained by the television camera,
A cross-correlation calculation with a linear FM signal whose frequency changes linearly with time is performed in real time, and this correlation output is swept onto the cathode ray tube surface to display a horizontal scanning direction spatial frequency spectrum image of the input image. An object of the present invention is to provide an image spatial frequency analyzer. To achieve this purpose, a FM signal generating device and a spectral image display device were provided, and a signal compression circuit was added to realize cross-correlation calculation between the analog value video signal and the linear FM signal.

Furthermore, a video signal sampling circuit was provided to form a no-signal portion in the video signal. Next, this invention will be specifically explained with reference to the drawings.

FIG. 1 is a block diagram of an embodiment of the image spatial frequency analyzer of the present invention.

The cross-correlation calculation operation of the optical correlator 101 is similar to the principle of the conventional device. In FIG. 1, the video signal of an input image 13 obtained by a television camera 14 that performs unidirectional or bidirectional horizontal scanning is removed by removing the horizontal and vertical synchronizing pulses, and at the same time removing every other video signal obtained in each horizontal scanning. The video signal is added to a video signal sampling circuit 18 which samples and outputs the video signal. At the same time, the television camera 14
The output is also sent to a monitor television 17 to display the input image being photographed. The output of the video signal sampling circuit 18 is sent to a square compression circuit 19 and subjected to square compression.
That is, the input video signal f(t
), the output signal of the circuit becomes f(x). This signal is applied as an external modulation signal to the amplitude modulated sinusoidal oscillator la, which constitutes the optical correlator described above. Next, a linear FM signal generator will be explained. The output of the sweep signal generation circuit 20 that generates a frequency triangular wave that periodically and linearly changes the frequency of the linear FM signal is as follows.
It is applied as an external modulation signal to an oscillator capable of frequency modulation (frequency modulation oscillator) 21. The frequency modulation oscillator 2
The linear FM signal generated in step 1 is applied as an external modulation signal to an amplitude modulated sinusoidal oscillator lb. At this time, if the degree of modulation is set so that the number of oscillations of the sine wave oscillator lb is overmodulated by 200% by the linear FM signal, it is equivalent to amplitude modulating the oscillation wave with a square compressed signal of the linear FM signal. become. In other words, the properties of trigonometric functions {■S cloud}
2 = Season (10 Sa)... Considering '51, the right side of the equation 5 is considered to be the external modulation wave at 100% amplitude modulation, and the left side of the same equation is the square of the external modulation wave at 200% overmodulation. .

Therefore, except that the frequency is doubled, setting a 200% overmodulation state is equivalent to applying external modulation with a square compressed wave. Now, the cross-correlation function Cs(t,B
) is expressed by the following formula. ■Formula 8 is the number of repetitions of the linear FM signal, k' is a constant for phase calculation, d is the beam axis, x is the axis of the ultrasound propagation direction in the ultrasonic optical modulator, Fs is the initial frequency of the linear FM signal, and Fc is Similarly, it represents the degree of frequency-time variation.

FIG. 2 shows a time chart when formula '6} is realized within the correlator. Figure a shows the sweep signal generation circuit 20.
Fig. 1b shows a linear FM signal whose frequency changes repeatedly at the period of the sweep signal. Figure c is a video signal obtained by sampling every other horizontal scan, and the video signal of the input image is included in the portion marked as signal range in the figure. The linear FM signal b and the same video signal c
are the velocities v in opposite directions within each ultrasonic light modulator.
It is progressing. Therefore, a cross-correlation calculation is performed at the portion where these two signals cross the light east at the same time. Figure d shows the results of this cross-correlation calculation. When the frequency shift degree Fc of the linear FM signal is small, the linear FM signal in Koto, which is undergoing correlation calculation with each video signal, can be regarded as a sine wave signal with a constant frequency. Therefore, each video signal undergoes a cross-correlation calculation with a sine wave signal having a different frequency, and the magnitude of the correlation output is determined by the degree of similarity between the video signal and the sine wave signal, that is, the sine wave signal included in the video signal. Indicates the component value of the wave signal frequency. As shown in c in the same figure, if the video signal contains a signal with a frequency fP, this signal and the linear F
Time t when the frequency fP part of the M signal overlaps in Koto d
The correlation output amplitude value shows the maximum value at P, and the waveform is 2P
vibrates at the frequency of At this time, as shown in the figure, the linear FM
If a DC part with a spatial distance of d/2 is created in the initial part of the signal, the repetition time t=0,
At T, 2r, . . . , the entire area of the video signal waveform, that is, the DC component, is always output. If this value is used as the standard value for device normalization, the component value of the video signal having only the frequency fP signal as described above will be equal to the standard value,
In addition, in the case of an input image that outputs various component values,
Each frequency component value can be defined as the ratio of the component value to the reference value. Next, since the correlation results between the video signal and each frequency sine wave in the above correlation calculation are discrete in terms of frequency, a method of configuring them into a frequency spectrum image will be described.

FIG. 3 shows how to construct an input image on a TV screen and its spatial frequency spectrum image. The magnitude of the component of the spectral image is expressed by brightness. The width Th of the video signal obtained in one horizontal scan is the repetition period Tf of the linear FM signal.
If it is small compared to , many video signals undergo correlation calculations for one cycle of linear FM signal.

Therefore, if the bright spots representing the spectral components are swept left and right on the spectrum display screen using the frequency sweep signal of the linear FM signal, the cross-correlation output with the video signal in various frequency bands of the linear FM signal, that is, the spectral components is displayed horizontally on the screen. In the example shown in the figure, the input image,
Both spectral images are displayed using bidirectional horizontal scanning.
This may be a unidirectional horizontal scan. Now, in the spectrum components displayed in the horizontal direction of the spectrum screen, each bright spot indicates a frequency component value included in every other horizontal scanning video signal of the input image. Therefore, if the rope point is further synchronized with the vertical synchronization pulse of the television camera 14 and swept from top to bottom on the screen, an input/output image corresponding to the horizontal scanning position as shown in the figure is obtained. Since the spectral image composed of this foot point becomes a Lissajous figure drawn by the frequency shift signal of the linear FM signal and the vertical sweep signal of the television camera 14, the repetition period of the two signals is slightly shifted from an integral multiple. This creates a fine mesh pattern that uniformly covers the output screen. Therefore, a horizontal scanning direction spatial spectral image of the input image can be displayed on the output screen, and this spectral image becomes a real-time display image according to changes in the input image. The present invention has the above-described configuration, and has the effect that spectrum analysis is possible even for moving images that can be photographed by the television camera 14.

Furthermore, if a television camera and photographing device are used in a fixed manner, it is also possible to measure the absolute value of the spatial frequency contained in the target image.In this case, if the image enlargement and reduction effects of the television camera lens system are used, More accurate measurements are also possible. Furthermore, the video signal to be input is not limited to the signal directly obtained from a television camera, but also video signals recorded on a videotape or the like can be used.

[Brief explanation of the drawing]

Fig. 1 is a diagram showing an embodiment of the present invention, Fig. 2 is a diagram showing a signal timing chart, Fig. 3 is a diagram showing the configuration of a spectral image, Fig. 4 is a diagram showing a conventional example, and Fig. 5 6 is a diagram showing bidirectional horizontal scanning of an image, and FIG. 6 is a diagram showing a signal timing chart. la and lb are sine wave oscillators, 2 is a missing number, 3a and 3b are broadband amplifiers, 4 is a Caesar, 5a and 5b are ultrasonic optical modulators, 6 is a photodetection filter, 7 is a photoelectric converter, and 8 is a Video signal amplifier, 9 is a video signal pulsing circuit, 1 is a synchronization pulse separation circuit, 1 is a horizontal synchronization time delay circuit, 12 is a mixer, 13 is an input image, 14 is a television camera, 15 is a bidirectional sweep oscillator , 16 is a television receiver, 17 is a monitor television, 18 is a video signal sampling circuit, 19 is a square compression circuit, 20 is a sweep signal generation circuit, 21 is a frequency modulation oscillator, 22 is a band filter, 23 is an output image display, 101
102 is a correlator, 102 is a Fourier transform optical system, 103 is a linear FM signal repetition generator, and ULMa and ULMb are ultrasonic optical modulators. Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6

Claims (1)

[Claims] 1 An image spatial frequency analyzer for displaying a spatial frequency spectrum image of an input image 13; A television camera 14 that outputs a composite video signal obtained by horizontally scanning the input image; a video signal sampling circuit 18 that separates pulses and samples the video signal obtained for each horizontal scan every other scan to create no-signal portions before and after the video signal; a video signal having the no-signal portion; a square compression circuit 19 that performs square compression; a linear FM signal repetition generator 103 that generates a linear FM signal at a constant repetition period;
A pair of amplitude modulated sine wave oscillators 1a and 1b, broadband amplifiers 3a and 3b whose input signals are the output signals of the sine wave oscillators, and whose ultrasonic propagation directions are opposite to each other and whose input signals are the output signals of the amplifiers. A Fourier transform optical system 102 having a pair of ultrasonic light modulators 5a and 5b, a laser 4 as a light source of the optical system, a filter 6 for detecting the output light of the optical system, and a photoelectric converter for the output light. a correlator 101 that receives the video signal having a no-signal portion and the linear FM signal and outputs a cross-correlation signal; The present invention is characterized by comprising an output image display 23 that receives the FM signal frequency sweep signal, the vertical synchronization sweep signal of the television camera, and the cross-correlation signal output from the correlator and displays a spatial frequency spectrum image. Image spatial frequency analyzer.